This is the accepted version of the article: Almécija, Sergio; Hammond, Ashley S.; Thompson, Nathan; [et al.]. «Fossil apes and human evolution». Science, Vol. 372, núm. 6542 (May 2021), art. eabb4363. DOI 10.1126/science.abb4363 This version is available at https://ddd.uab.cat/record/240044 under the terms of the license 1 Fossil apes and human evolution 2 Sergio Almécija1,2,3,*, Ashley S. Hammond1,2, Nathan E. Thompson4, Kelsey D. Pugh1,2, 3 Salvador Moyà-Solà3,5,6, David M. Alba3 4 1 5 Division of Anthropology, American Museum of Natural History (AMNH). New York, NY 10024, 6 USA 2 7 New York Consortium in Evolutionary Primatology at AMNH. New York, NY 10024, USA 3 8 Institut Català de Paleontologia Miquel Crusafont (ICP), Universitat Autònoma de Barcelona. 9 08193 Cerdanyola del Vallès, Barcelona, Spain 4 10 Department of Anatomy, NYIT College of Osteopathic Medicine. Old Westbury, NY 11568 USA 5 11 Institució Catalana de Recerca i Estudis Avançats (ICREA). 08010 Barcelona, Spain 6 12 Unitat d’Antropologia Biològica, Departament de Biologia Animal, Biologia Vegetal i Ecologia, 13 Universitat Autònoma de Barcelona. 08193 Cerdanyola del Vallès, Barcelona, Spain 14 *Corresponding author. Email: salmecija@amnh.org 15 16 Single-sentence summary 17 Humans are one of the five surviving hominoid lineages, a decimated remainder of an ancient 18 radiation whose fossil members are essential to understanding human origins. 19 20 Abstract 21 Humans diverged from apes (chimpanzees, specifically) toward the end of the Miocene ~9.3–6.5 22 million years ago. Understanding the origins of the human lineage (hominins) requires 23 reconstructing the morphology, behavior, and environment of the chimpanzee–human last 24 common ancestor. Modern hominoids (i.e., humans and apes) share multiple features (e.g., an 25 orthograde body plan facilitating upright positional behaviors). However, the fossil record 26 indicates that living hominoids constitute narrow representatives of an ancient radiation of more 27 widely distributed, diverse species, none of which exhibit the entire suite of locomotor adaptations 28 present in the extant relatives. Hence, some modern ape similarities might have evolved in parallel 29 in response to similar selection pressures. Current evidence suggests that hominins originated in 30 Africa from Miocene ape ancestors unlike any living species. 31 1 32 33 In 1871, Darwin (1) speculated that humans originated in Africa based on the anatomical 34 similarities with African apes (gorillas and chimpanzees) identified by Huxley (2). However, 35 Darwin urged caution until more fossils became available—the European Dryopithecus was the 36 only recognized fossil ape at the time (3). After 150 years of continuous discoveries, essential 37 information about human origins remains elusive due to debates surrounding the interpretation of 38 fossil apes (Figs. 1, 2). 39 40 Genomic data indicate that humans and chimpanzees are sister lineages (“hominins” and “panins,” 41 respectively; Table 1) that diverged from a “last common ancestor” (LCA) toward the end of the 42 Miocene, ~9.3–6.5 million years ago (Ma) (4, 5). All extant hominoids (apes and humans) are 43 characterized by the lack of an external tail, high joint mobility (e.g., elbow, wrist, hip), and the 44 possession of an “orthograde” (upright) body plan —as opposed to the more primitive, 45 “pronograde” body plan of other anthropoids and most other mammals (Fig. 2). These body plans 46 are associated with two different types of positional (postural and locomotor) behaviors: 47 pronograde behaviors, taking place on nearly horizontal supports with the trunk held roughly 48 horizontally; and orthograde (or “antipronograde”) behaviors, with the torso positioned vertically 49 (6, 7). Extant ape features also include enhanced joint mobilitiy, long forelimbs relative to 50 hindlimbs, and (except gorillas) long hands with high-to-very-high finger curvature (8-10). The 51 orthograde body plan is generally interpreted as a suspensory adaptation (11, 12), or as an 52 adaptation for vertical climbing subsequently co-opted for suspension (13). 53 54 Based on similarities between chimpanzees and gorillas, a prevalent evolutionary model argues 55 that African apes represent “living fossils” and that knuckle-walking chimpanzees closely reflect 56 the morphology and behavior of the Pan–Homo LCA—the “starting point” of human evolution 57 (14, 15). This working paradigm also postulates that modern African apes occupy the same habitats 58 as their ancestors (16) (Fig. 1). This assumption is based on a classical scenario that situates 59 hominin origins in East Africa, due to environmental changes following the rifting of East African 60 Rift Valley during the Miocene (17). For some, a chimpanzee-like Pan–Homo LCA could imply 61 also that all extant ape locomotor adaptations were inherited from a modern ape-like ancestor (18). 62 However, the fossil record denotes a more complex picture: Miocene apes often display mosaic 2 63 morphologies, and even those interpreted as crown hominoids do not exhibit all the features 64 present in living apes (19) (Fig. 3). 65 66 The Pan-like LCA model builds on the “East side story” of hominin origins (17), a seriously- 67 challenged scenario. First, it is grounded in the living ape geographic distribution, which may not 68 match that at the time of the Pan–Homo split (Fig. 1). Second, the model relies on an outdated 69 account of the fossil record (from the 1980s), when the earliest known hominin (Australopithecus 70 afarensis) was recorded in East Africa, and no possible fossil gorillas and chimpanzees were 71 known (17). Subsequent fossil discoveries are incompatible with such a narrative: 72 Australopithecus remains from Chad indicate that early hominins were living ~2,500 kilometers 73 west of the Rift ~3.5 Ma (20). Furthermore, if Sahelanthropus is a hominin, it would push back 74 the human lineage presence in northcentral Africa to ~7 Ma (21). Moreover, continued fieldwork 75 efforts in less explored areas have shown that hominoids lived across Afro-Arabia during the 76 Miocene (22-25). In addition, remains of putative hominines have been found in East Africa (26, 77 27), perhaps even in Europe (28, 29). Finally, paleoenvironmental reconstructions for late Miocene 78 apes and hominins suggest the Pan-Homo LCA inhabited woodlands, not tropical rainforests (30- 79 33). 80 81 Current debates about the transition from an ape into a bipedal hominin are centered on the 82 morphological and locomotor reconstruction of the Pan–Homo LCA, as well as its 83 paleobiogeography. Discrepancies are caused by conflicting evolutionary signals among living 84 and fossil hominoids—indicating rampant “homoplasy” (independent evolution causing “false 85 homology”)—and further complicated by the highly incomplete and fragmentary nature of the 86 hominoid fossil record. This review argues that, in spite of the limitations, the information 87 provided by fossil apes is essential to inform evolutionary scenarios of human origins. 88 89 90 Evidence as to humans’ place in nature 91 Humans’ inner primate 92 Since Linnaeus established modern taxonomy in 1758 (34) and until the 1960s, morphological 93 similarity was the main basis for classifying organisms. Linnaeus included modern humans (Homo 3 94 sapiens) within the order Primates, but it was not until 1863 that Huxley provided the first 95 systematic review of differences and similarities between humans and apes (2). Imagining himself 96 as a “scientific Saturnian” Huxley stated: “The structural differences between Man and the Man- 97 like apes certainly justify our regarding him as constituting a family apart from them; though, 98 inasmuch as he differs less from them than they do from other families of the same order, there 99 can be no justification for placing him in a distinct order” (2, p. 104). Huxley’s work was motivated 100 by widespread claims (e.g., Cuvier, Owen) that humans’ “uniqueness” warranted their placement 101 in a separate order. Darwin concurred with Huxley that humans should be classified in their own 102 family within primates (1). 103 104 We now know that most “human features” are primitive traits inherited from primate (e.g., 105 trichromatic stereoscopic vision, manual grasping) or earlier (e.g., five digits) ancestors (35). Even 106 humans’ uniquely large brains and delayed maturation are framed within a primate trend of 107 increased encephalization and slower life history compared with other mammals (35, 36). Some 108 differences in brain size may partly reflect a neocortex enlargement related to enhanced visual and 109 grasping abilities (37). Like extant great apes, humans display larger body size, larger relative 110 brain size, a slower life history profile, and more elaborated cognitive abilities than other primates 111 (hylobatids included) (36). However, modern humans are extreme outliers in terms of delayed 112 maturation, encephalization, advanced cognition and manual dexterity, ultimately leading to 113 symbolic language and technology (38). 114 115 Anatomically, only two adaptive complexes represent synapomorphies present in all hominins: the 116 loss of the canine honing complex and features related to habitual bipedalism (33, 39). Most 117 anthropoids possess large and sexually dimorphic canines coupled with body size differences 118 between males and females, reflecting levels of agonistic behavior and sociosexual structure (40). 119 The fossil record indicates that there was a reduction in canine height, leading to the loss of the 120 honing complex in early hominins (41). Habitual bipedalism is reflected in several traits across the 121 body (e.g., foramen magnum position/orientation; pelvic, lower back and lower limb morphology), 122 present (or inferred) in the earliest hominins (21, 33, 42). 123 4 124 Darwin linked origin of bipedalism with an adaptive complex related to freeing the hands from 125 locomotion to use and make tools (replacing large canines), leading to a reciprocal feedback loop 126 involving brain size, cognition, culture, and eventually civilization (1). Multiple variants in the 127 order of these events have been advocated, with the freeing of the hands alternatively linked to 128 tools (43), food acquisition and carrying (15), or provisioning within a monogamous social 129 structure (44), to name a few. There is general agreement that canine reduction (including social 130 structure changes), enhanced manipulative capabilities and bipedalism were interrelated during 131 human evolution. However, determining the order of events and their causality requires 132 reconstructing the ape–human last common ancestor (LCA) from which hominins originated. 133 Darwin also speculated that humans and modern African ape ancestors originated in Africa (1), 134 based on the anatomical similarities identified by Huxley, and his own observations that many 135 living mammals are closely related to extinct species of the same region. However, given the 136 limited ape fossil record then, he concluded that it was “useless to speculate on this subject” (1, p. 137 199). Using the French Dryopithecus to calibrate his “clock,” Darwin concluded that humans likely 138 diverged as early as the Eocene, and warned against “the error of supposing that the early 139 progenitor of the whole Simian stock, including man, was identical with, or even closely 140 resembled, any existing ape or monkey” (1, p. 199). These ideas inaugurated a century of 141 discussions about human’s place in nature. 142 143 Reaching the “extant” consensus 144 Until the 1950s, the geographic origin of hominins was disputed between Africa, Asia, and Europe. 145 Following the publication of Darwin’s On the Origin of Species (45), Haeckel predicted that the 146 “missing link” (dubbed “Pithecanthropus,” the “ape-man”) would be found in Asia (46). This idea 147 led to Dubois’ 1891 discovery of Homo erectus in Indonesia (47). In 1925, Dart published the 148 discovery of Australopithecus africanus, “the man-ape from South Africa” (48). However, the 149 scientific community still focused on Europe due to the Piltdown “fossils,” until exposed as a hoax 150 (49). Asia remained a “mother continent” contender due to the “man-like ape” Ramapithecus, 151 discovered in the Indian Siwaliks (50). 152 153 During this time, the relationships of humans to other primates were highly contentious. Most 154 authors advocated an ancient divergence of humans from apes (51, 52), or favored a closer 5 155 relationship to the great apes than to the lesser apes (53, 54). A few proposed that humans were 156 more closely related to one or both of the African apes (55, 56), although these views were not 157 widely accepted (57). These alternative phylogenetic hypotheses heavily impacted reconstructions 158 of the LCA. Some (e.g., Schultz, Straus) advocated for a “generalized” ape ancestor (52), while 159 others relied on extant hominoid models. Notably, Keith developed a scenario in which a 160 “hylobatian” brachiating stage preceded an African ape-like creature: a knuckle-walking 161 “troglodytian” phase immediately preceding bipedalism (11). Focused on Keith’s “hylobatian” 162 stage, Morton proposed that the “vertically suspended posture” of a small-bodied hylobatid-like 163 ancestor caused the erect posture of human bipedalism (12). Gregory, another prominent 164 “brachiationist,” supported similar views (53). Morton argued that knuckle walking did not 165 represent an intermediate stage preceding bipedalism, but a reversion toward quadrupedalism in 166 large-bodied apes specialized for brachiation. Then, “brachiation” was used for any locomotion in 167 which the body was suspended by the hands. Currently it refers to the pendulum-like arm-swinging 168 locomotion of hylobatids (6). 169 170 By the 1960s, the Leakeys’ discoveries in Tanzania, [e.g., Paranthropus boisei (58), Homo habilis 171 (59)], reinforced the relevance of Africa in human evolution, which became established as the 172 “mother continent” with the Australopithecus afarensis discoveries during the 1970s (60, 61). 173 LCA models still centered on the available fossil apes (mostly represented by jaw fragments and 174 isolated teeth), found after decades of paleontological fieldwork in Africa and Eurasia. In 1965, 175 Simons and Pilbeam (62) revised and organized available Miocene apes in three genera: 176 Dryopithecus, Gigantopithecus and Ramapithecus. The genus Sivapithecus was included in 177 Dryopithecus, considered the ancestor of African apes, whereas Ramapithecus was considered 178 ancestral to humans based on its short face (and inferred small canines) (63). Leakey (64) and 179 others agreed with Simons and Pilbeam that humans belong to their own family (Hominidae, or 180 “hominids”), whereas great apes would belong to a distinct family (Pongidae, or “pongids”). He 181 also agreed that Ramapithecus was an Asian early human ancestor. However, Leakey proposed 182 reserving the genus Sivapithecus for the “Asian dryopithecines,” and claimed that the human 183 lineage could be traced back to, at least, the middle Miocene of Africa with Kenyapithecus wickeri 184 (~14 Ma). 185 6 186 Two major “revolutions” in the study of evolutionary relationships started in the 1960s. First, a 187 series of studies jumpstarted the field of molecular anthropology: Blood protein comparisons by 188 Zuckerkandl et al. (65) and Goodman (66) found that some great apes—gorillas and 189 chimpanzees—were more closely related to humans than to orangutans. Sarich and Wilson 190 developed an “immunological molecular clock” and concluded that African apes and humans share 191 a common ancestor as recent as ~5 Ma (67). These results led to decades-long debates regarding 192 the African ape–human split. For example, Washburn resurrected extant African apes as ancestral 193 models in human evolution, proposing knuckle walking as the precursor of terrestrial bipedalism 194 (68). In contrast, paleontologists argued that the molecular clock was inaccurate because of the 195 much older age of the purported human ancestors Kenyapithecus and Ramapithecus (69). Second, 196 Hennigian cladistics (“phylogenetic systematics”)—which only recognizes “synapomorphies” 197 (shared derived features) as informative for reconstructing phylogeny (70)—became slowly 198 implemented in anthropology by the mid-1970s (71). 199 200 In the 1970–1980s, the relationships among gorillas, chimpanzees and humans were still disputed. 201 Chromosomal comparisons (72), DNA hybridization (73), and hemoglobin sequencing (74) 202 supported a closer relationship between chimpanzees and humans, whereas morphology-based 203 cladistics recovered gorilla–chimpanzee as monophyletic (75). In the late 1980s, the first single- 204 locus DNA sequencing studies (76) followed in the 1990s with multiple loci analyses finally 205 resolved the “trichotomy” (77). Current genomic evidence indicates that humans are more closely 206 related to chimpanzees (5), having diverged ~9.3–6.5 Ma (4). Ever since “the molecular 207 revolution,” the perceived relevance of fossil apes in human evolution has been in jeopardy. 208 209 African apes as time machines? 210 Extant African apes have been considered ancestral models since Keith’s “troglodytian” stage in 211 the 1920s (11), and especially since the 1960s, with updated hypotheses inspired by the “molecular 212 revolution” (68, 78) and field discoveries on chimpanzee behavior by Goodall (79). Louis Leakey 213 played a central role in promoting Goodall’s pioneering research (subsequently fostering Fossey’s 214 in gorillas and Galdikas’s in orangutans). Currently, a prominent paradigm proposes that 215 chimpanzees represent “living fossils” closely depicting the Pan–Homo LCA (14, 16). This model 216 combines molecular data with the anachronistic view that Gorilla and Pan are morphologically 7 217 similar (75). Under these assumptions, knuckle walking—once used to defend African ape 218 monophyly (80)—is used to argue that African apes are morphologically “conservative” and only 219 display size-related differences (14). This model contends that gorillas are allometrically enlarged 220 chimps and that chimpanzees [or bonobos (78)] constitute a suitable model for the Pan–Homo 221 LCA, perhaps even the hominine or hominid LCAs (14). This narrative also incorporates the 222 paleobiogeographic assumption that African apes likely occupy the same habitats as their 223 ancestors: Without new selection pressures, there was no need for evolution. 224 225 If hominins originated from a chimpanzee-like LCA, human bipedalism must have evolved from 226 knuckle walking (15)—a functional compromise enabling terrestrial travel while retaining 227 climbing adaptations (80). Under this view, bipedal hominins originated from an ancestor that was 228 already terrestrial while traveling. These conclusions are logical from a “top-down” perspective, 229 based on the evidence provided by extant hominoids and early hominins. However, a fully- 230 informed theory of hominin origins must also apply a “bottom-up” approach (81, 82), from the 231 perspective of extinct apes preceding the Pan–Homo split. It is also essential to clarify whether 232 chimpanzees represent a good ancestral model for the Pan–Homo LCA. Unfortunately, the view 233 from the bottom is blurry. 234 235 236 The tangled branches of ape evolution 237 The fossil ape dilemma: Homoplasy and mosaic evolution 238 With more than 50 hominoid genera and a broad geographic distribution (Fig. 1), the Miocene has 239 been dubbed “The real planet of the apes” (83). Besides their fragmentary nature, a persistent 240 challenge is understanding the phylogenetic relationships among fossil apes exhibiting mosaics of 241 primitive and derived features with no modern analogs. The Asian Miocene ape Sivapithecus best 242 exemplifies this complexity. Discoveries during the 1970s and 1980s, including a facial skeleton 243 (84), clarified that Ramapithecus is a junior synonym of Sivapithecus, which is likely related to 244 orangutans (85). However, two Sivapithecus humeri show a primitive (pronograde-related) 245 morphology, calling into question the close phylogenetic link with Pongo inferred from facial 246 similarities (86). 247 8 248 The root of this “Sivapithecus dilemma” (18) is identifying where “phylogenetic signal” is best 249 captured in hominoids: the postcranium or the cranium? The former implies that a Pongo-like face 250 evolved independently twice; the latter that some postcranial similarities among living apes 251 evolved more than once. Both hypotheses highlight the phylogenetic noise that homoplasy 252 introduces in phylogenetic inference. Indeed, several studies have found that homoplasy similarly 253 affects both anatomical areas (87). The conclusion that Sivapithecus is not a pongine relies on the 254 assumption that suspensory adaptations and other orthograde-related features present in living 255 hominoids were inherited from their LCA (18). However, this is contradicted by differences among 256 living apes [e.g., forelimb and hand anatomy, degree of limb elongation, hip abduction capability 257 (8, 9, 19, 80, 88-91)]. These studies concluded that apparent similarities could represent 258 independently evolved biomechanical solutions to similar locomotor selection pressures. For 259 instance, hand length “similarities” among living apes result from different combinations of 260 metacarpal and/or phalangeal elongation in each extant genus (9). 261 262 Parallel evolution—homoplasy among closely related taxa due to shared genetic and 263 developmental pathways—could explain some postcranial similarities related to suspensory 264 behaviors among extant apes (80). Compared with convergences among distantly-related taxa, 265 parallelisms are more subtle and difficult to detect, and readily evolve when similar selection 266 pressures appear. Within extant primates, suspensory adaptions evolved independently in atelines 267 and between hylobatids and great apes (8, 80, 88, 91, 92). When the hominoid fossil record is 268 added, independent evolution of suspensory adaptations has been inferred too for orangutans, 269 chimpanzees, and some extinct lineages (9, 89, 93, 94). Knuckle walking has also been proposed 270 to have different origins in gorillas and chimpanzees (80, 93, 95). As for suspension, the pre- 271 existence of an orthograde body plan, vertical climbing, and general arboreal heritage could have 272 facilitated the independent evolution of knuckle walking to circumvent similar biomechanical 273 demands during terrestrial quadrupedalism, while preserving a powerful grasping hand suitable 274 for arboreal locomotion (9). 275 276 The possibility of parallelisms indicates that ancestral nodes in the hominoid evolutionary tree— 277 including the Pan–Homo LCA—cannot be readily inferred without incorporating fossils. In 278 addition, fossils from “known” evolutionary lineages are commonly used to calibrate molecular 9 279 clocks despite being subject to considerable uncertainty (4). Even worse, relatively complete fossil 280 apes undisputedly assigned to early members of the gorilla and chimpanzee lineages remain to be 281 found. 282 283 Counting crowns: The case of the European Miocene apes 284 Sivapithecus and other fossil Asian great apes (e.g., Khoratpithecus, Ankarapithecus, 285 Lufengpithecus) are generally considered pongines (Fig. 3) based on derived craniodental traits 286 shared with Pongo (94, 96-98), although alternative views exist, particularly for Lufengpithecus 287 (99). In contrast, the phylogenetic position of apes from the African early (e.g., Ekembo, 288 Morotopithecus) and middle Miocene (Kenyapithecus, Nacholapithecus, Equatorius) remains 289 very controversial. Like Sivapithecus, they exhibit only some modern hominoid features 290 superimposed onto a primitive-looking pronograde (“monkey-like”) body plan (Fig. 2). Some 291 authors interpret this mosaicism as indicating that most Miocene apes do not belong within the 292 crown hominoid radiation and, thus, are irrelevant to reconstructions of the Pan–Homo LCA (14). 293 This is likely the case for early Miocene African taxa. However, the vertebrae of Morotopithecus 294 [~20 Ma (100) or ~17 Ma (101)] display orthogrady-related features absent from other stem 295 hominoids—indicating either a closer relationship with crown hominoids or an independent 296 evolution of orthogrady (102). In turn, Kenyapithecus and Nacholapithecus are commonly 297 regarded as preceding the pongine-hominine split due to the possession of some modern hominid 298 craniodental synapomorphies combined with a more primitive postcranium than in living great 299 apes (94, 103). This raises the question: Can some Miocene apes belong to the crown hominid 300 clade despite lacking many of the features shared by extant great apes? 301 302 The large-bodied apes from the middle-to-late Miocene of Europe are at the center of discussions 303 about great ape and human evolution (19, 28, 94, 104, 105). Named after Dryopithecus (3), they 304 are generally distinguished as a subfamily (Dryopithecinae) (94) or tribe (Dryopithecini) (28). 305 However, it is unclear if they constitute a monophyletic group or a paraphyletic assemblage of 306 stem and crown hominoids (94). Thus, we refer to them informally as “dryopiths.” These apes are 307 dentally conservative, but each genus exhibits different cranial and postcranial morphology. The 308 dryopith fossil record includes the oldest skeletons consistently exhibiting postcranial features of 309 living hominoids (orthograde body plan and/or long and more curved digits). Dryopithecus (~12– 10 310 11 Ma) is known from craniodental remains and isolated postcranials too scarce to reconstruct its 311 overall anatomy (106). In contrast, Pierolapithecus (~12 Ma) is represented by a cranium with 312 associated partial skeleton (19). Cranially a great ape, its rib, clavicle, lumbar, and wrist 313 morphologies are unambiguous evidence of an orthograde body plan. Yet, unlike chimpanzees and 314 orangutans (but similar to gorillas), Pierolapithecus lacks specialized below-branch suspensory 315 adaptations [see discussion in (10)]. The recently described Danuvius (~11.6 Ma, Germany), and 316 the slightly younger (~10–9 Ma) Hispanopithecus (Spain) (105) and Rudapithecus (Hungary) (28) 317 represent the oldest record of specialized below-branch suspensory adaptations (e.g., long and 318 strongly curved phalanges; Fig. 2). Danuvius has also been argued to show adaptations to habitual 319 bipedalism (but see below). 320 321 The different mosaic morphology exhibited by each dryopith genus is a major challenge for 322 deciphering their phylogenetic relationships (Fig. 3). Current competing phylogenetic hypotheses 323 consider dryopiths as stem hominoids (107, 108), stem hominids (94, 96, 109) or crown hominids 324 closer to either pongines (105), hominines (28), or even hominins (29, 110). However, recent 325 phylogenetic analyses of apes recovered dryopiths as stem hominids (97, 109), perhaps except 326 Ouranopithecus (~9–8 Ma) and Graecopithecus (~7 Ma) (97). Ouranopithecus has been 327 interpreted by some as a stem hominine, or even as a crown member more closely related to the 328 gorilla or human lineages (110). Graecopithecus has also been advocated as a hominin (29), 329 although the fragmentary available material hinders evaluation of this hypothesis. Such contrasting 330 views about dryopiths stem from their incomplete and fragmentary fossil record coupled with 331 pervasive homoplasy. However, as these factors are equal for all researchers, their different 332 conclusions must also relate to analytical differences (e.g., taxonomy, sampling, polymorphic and 333 continuous trait treatment). The root of the conflict is the striking differences in subjective 334 definition and scoring of complex morphologies (e.g., “incipient supraorbital torus”). 335 336 Paleobiogeography of the African ape and human clade 337 150 years after Darwin speculated that modern African ape and human ancestors originated in 338 Africa, possible hominins have been found as far back as the latest Miocene of Africa (21, 33, 339 111): Sahelanthropus (~7 Ma), Orrorin (~6 Ma), and Ardipithecus kadabba (~5.8–5.2 Ma). 340 However, others question the feasibility of identifying the earliest hominins among the diverse 11 341 Miocene apes (96, 112). Puzzlingly, despite some claims based on scarce remains (113-115), 342 ancient representatives of the gorilla and chimpanzee lineages remain elusive. Some apes from the 343 African late Miocene—Chororapithecus (26), Nakalipithecus (27), and Samburupithecus (116)— 344 have been interpreted as hominines, but the available fragmentary remains preclude a conclusive 345 assessment. Furthermore, Samburupithecus is likely a late occurring stem hominoid (97, 117). 346 347 During the middle Miocene (~16.5–14 Ma), apes are first found “out of Africa.” These are the 348 genera Kenyapithecus (Turkey) and Griphopithecus (Turkey and central Europe). We informally 349 refer to them as the “kenyapiths” because there is no consensus on their relationships (28, 94, 118). 350 Kenyapiths indicate that putative stem hominids are first recorded in Eurasia and Africa before the 351 earliest record of both European dryopiths and Asian pongines at ~12.5 Ma (94). 352 Paleobiogeographical and paleontological data suggest that kenyapiths dispersed from Africa into 353 Eurasia as one of the multiple catarrhine intercontinental dispersal events occurred during the 354 Miocene (e.g., hylobatids, pliopithecoids) (83, 94). While some competing evolutionary scenarios 355 agree that kenyapiths gave rise to dryopiths in Europe, the phylogenetic and geographic origin of 356 hominines remains contentious (28, 94). 357 358 If dryopiths are stem hominids, they could either be close to the crown group or constitute an 359 evolutionary dead-end, an independent “experiment” not directly related to either pongines or 360 hominines. Alternatively, dryopiths might be crown hominids more closely related to one of these 361 groups. If dryopiths are hominines, this implies that the latter could have originated in Europe and 362 subsequently dispersed “back to Africa” during the late Miocene (28, 29, 83). This would coincide 363 with vegetation structure changes caused by a trend of increased cooling and seasonality (32) that 364 ultimately drove European apes to extinction [or back to Africa (28)]. In this scenario, hominines 365 and pongines would be vicariant groups that originally evolved in Europe and Asia, respectively, 366 from early kenyapith ancestors. Given the suspensory specializations of late Miocene dryopiths 367 (Hispanopithecus and Rudapithecus), if modern African apes originated from these forms, this 368 scenario involves that the hominine ancestor could have been more reliant on suspension than 369 living chimpanzees or gorillas. The claim that hominines originated outside of Africa may be 370 justified by cladistic analyses recovering dryopiths as stem hominines, but not based on the lack 371 of late Miocene great apes in Africa because fossils from this critical time period have been 12 372 discovered (~13–7 Ma) (Fig. 3). Both molecular and paleontological evidence (e.g., Sivapithecus) 373 situate the pongine-hominine divergence within the middle Miocene. Hence, the debate cannot be 374 settled without more conclusively resolving the phylogenetic relationships of middle Miocene 375 dryopiths. 376 377 An alternative scenario proposes a vicariant divergence for hominines and pongines from 378 kenyapith ancestors, but favors the origin of hominines in Africa (94, 119). It argues for a second 379 vicariant event between European dryopiths and Asian pongines soon after the kenyapith dispersal 380 into Eurasia. Cladistically, dryopiths would be pongines, but would share none of the currently- 381 recognized pongine autapomorphies, evolved after the second vicariant event. This scenario is 382 difficult to test, but it would be consistent with the apparent absence of clear pongine 383 synapomorphies in Lufengpithecus (99) and the more derived nasoalveolar morphology of 384 Nacholapithecus (103) compared with some dryopiths (106). However, it would imply even higher 385 levels of homoplasy—including the independent acquisition of an orthograde body plan in Africa 386 and Eurasia from pronograde kenyapith ancestors. 387 388 A third possibility is that none of the taxa discussed above are closely related to the African ape 389 and human clade (107). Under this view, bona fide extinct non-hominin hominines have yet to be 390 found in largely unexplored regions of Africa—explaining the virtual lack of a gorilla and 391 chimpanzee fossil record. According to Pilbeam, paleoanthropologists could be “like the drunk 392 looking for his keys under the lamppost where it was light rather than where he had dropped them, 393 working with what we had rather than asking whether or not that was adequate” (108, pp. 155- 394 156). Africa is a huge continent and most paleontological discoveries are concentrated in a small 395 portion of it. The greatest challenge is finding hominoid-bearing Mio-Pliocene sites outside East 396 and South Africa, even though we know they exist (20-22). Besides insufficient sampling effort, 397 this is hindered by numerous impediments to fieldwork in most of Africa, including geopolitical 398 conflicts, restricted land use development, lack of suitable outcrops (due to extensive vegetation 399 cover), and taphonomic factors [tropical forests do not favor fossil preservation (120)]. 400 13 401 A Miocene view of (Miocene) hominin origins 402 Evolution in motion 403 The decades-long feud regarding arboreality and bipedalism in Australopithecus afarensis 404 exemplifies the complexity of inferring function from anatomy. “Totalist” functional 405 morphologists rely on a species’ “total morphological pattern” (121) to infer its locomotor 406 repertoire. Totalists see a bipedal early hominin with some ape-like retentions (e.g., curved fingers) 407 pointing to continued use of the trees, and that certain not-yet-human-like features (e.g., hip) 408 indicate a different type of bipedalism (122). Instead, “directionalists”—for whom functional 409 inferences are only possible for derived traits evolved for a specific function—focus exclusively 410 on bipedal adaptations (123). Totalist and directionalist interpretations of the fossil record differ 411 in the “adaptive significance” attributed to primitive features, which result in different behavioral 412 reconstructions. Two other related factors further complicate locomotor inferences in extinct 413 species: First, different positional behaviors have similar mechanical demands [e.g., bipedalism, 414 quadrupedalism and some types of climbing (39)]. Second, pre-existing morphofunctional 415 complexes originally selected to fulfill a particular function (adaptations) can be subsequently co- 416 opted for a new role (exaptations). 417 418 The mosaic nature of hominoid morphological evolution makes the functional reconstruction of 419 fossil apes especially challenging, as recently exemplified by Danuvius (104): It was described as 420 possessing long and curved fingers, a long and flexible vertebral column, hip and knee joints 421 indicative of extended postures, and an ankle configuration aligning the foot perpendicular to the 422 long axis of the tibia. Such a combination of features was functionally interpreted as indicating 423 below-branch suspension combined with above-branch bipedalism. However, a critique to the 424 original study concluded that the morphological affinities of Danuvius with modern great apes 425 support a positional repertoire including orthogrady and suspension, but not bipedalism (124). Part 426 of the “problem” with the original interpretation is that it infers a derived locomotor behavior— 427 bipedalism—from primitive features that are also functionally related to quadrupedalism. For 428 instance, the inferred “long-back” morphology of Danuvius is characteristic of most quadrupedal 429 monkeys and other Miocene apes (125), denoting the lack of trunk specialization seen in extant 430 great apes. The Danuvius femoral head joint, being (primitively) posterosuperiorly expanded 431 (126), is consistent with flexed quadrupedal hip postures that are not used during human-like 14 432 bipedalism. In addition, the distal tibia configuration of Danuvius is shared with Ekembo and 433 cercopithecoids (104), thus being likely plesiomorphic and not unique to bipeds. When the 434 primitive and derived features of Danuvius are considered, a totalist would argue that it combined 435 high degrees of plesiomorphic quadrupedal locomotion with novel (suspensory) behaviors, 436 whereas a directionalist would downplay the primitive features in favor of the newly derived 437 adaptive traits (i.e., suspension). 438 439 The late Miocene Oreopithecus (~7 Ma, Italy) is another example of conflicting phylogenetic and 440 functional signals. Phylogenetic interpretations of Oreopithecus include cercopithecoid, stem 441 hominoid and hominid (even hominin) status (127). However, current phylogenetic analyses 442 suggest that Oreopithecus could represent a late occurring stem hominoid (97, 128), with 443 postcranial adaptations to alternative types of orthogrady, such as forelimb-dominated behaviors 444 (129) and terrestrial bipedalism (130). Even if not directly related to hominins (or modern 445 hominoids), the locomotor adaptations of Oreopithecus—and other Miocene apes—are worthy of 446 further research to understand the selection pressures that led to the (independent) emergence of 447 modern hominoid positional behaviors. 448 449 To distinguish true locomotor adaptations from exaptations, current research efforts focus on 450 plastic “ecophenotypic” traits—potentially denoting how fossil hominoids were actually moving. 451 Bone is a living tissue, and growth is expected to occur in predictable ways that reflect loading 452 patterns throughout life (131). Thus, cross-sectional and trabecular bone properties and their links 453 to behavior are widely investigated (132, 133). Yet, experimental studies indicate that internal 454 bone morphology does not necessarily match stereotypical loading patterns (134). Ample evidence 455 suggests that irregular loading, including low-magnitude, can be more osteogenically potent than 456 stereotypical loading (135). This may bias interpretations of individual fossils with a species- 457 atypical loading pattern during life (e.g., due to an injury). Bone (re)modeling also does not 458 consistently occur in response to changes in loading pattern: It can occur in ways that detract 459 from—rather than enhance—function (136), and may manifest differentially across the skeleton 460 (137). Incongruence also exists between actual bone performance and expectations based on 461 aspects of internal morphology (138). Finally, there is a strong genetic component to the 462 responsiveness of bone (re)modeling to loading (136), which is largely unknown for most species. 15 463 The confidence with which internal bone structures can be used to retrodict behavior in fossil 464 species remains a work in progress. 465 466 Before bipedalism 467 Competing hypotheses about the locomotor behavior immediately preceding hominin bipedalism 468 include terrestrial knuckle walking (15), palmigrade quadrupedalism (93), and different types of 469 arboreal (orthograde) behaviors such as climbing and suspension (7), vertical climbing (139), or 470 arboreal bipedalism and suspension (104, 140). Miocene great apes can enlighten this question by 471 helping to identify the polarity of evolutionary change preceding the Pan–Homo divergence (81, 472 82). For instance, if Pierolapithecus is interpreted as an orthograde ape without specific suspensory 473 adaptations but retaining quadrupedal adaptations [see alternatives in (10)], the orthograde body 474 plan and ulnocarpal contact loss could be interpreted as an adaptation to vertical climbing, 475 subsequently co-opted for suspension (19). Similarly, habitual bipedalism might have directly 476 evolved from other orthograde behaviors without an intermediate stage of advanced suspension or 477 specialized knuckle walking. Hence, Pierolapithecus complements previous hypotheses that 478 biomechanical aspects of the lower limb during quadrupedalism and vertical climbing could be 479 functionally “pre-adaptive” for bipedalism (39, 139). 480 481 A holistic view indicates that the Pan–Homo LCA was a Miocene ape with extant great ape-like 482 cognitive abilities, likely possessing a complex social structure and tool traditions (36, 38, 141). 483 This ape would exhibit some degree of body size and canine sexual dimorphism (with large honing 484 male canines) (15), indicating a polygynous sociosexual system (40). Based on Miocene apes and 485 earliest hominins, it is also likely that the Pan–Homo LCA was orthograde and proficient at vertical 486 climbing [see alternative interpretation based on Ardipithecus (33, 93)], but not necessarily at 487 specialized below-branch suspension or knuckle walking (9, 33). Chimpanzees seem to retain the 488 Pan–Homo LCA plesiomorphic condition in some regards [e.g., brain and body size (38), vertebral 489 counts (125), foot morphology (142)]. However, in others [e.g., interlimb (93), hand (9), pelvis 490 (143) length proportions; femur morphology (89)] early hominins are more similar to generalized 491 Miocene apes. These results further reinforce the idea that functional aspects of other locomotor 492 types were co-opted for bipedalism during hominin origins. 493 16 494 The “East Side Story” scenario links the divergence of chimpanzees and humans to the rifting of 495 East Africa, which would have triggered a vicariant speciation event from the ancestral Pan–Homo 496 LCA (17). Chimpanzees would have remained “frozen in time” in their ancestral tropical forest 497 environment, while humans would be the descendants of the group “left behind” on the east side 498 of the Rift. Major climate and landscape changes would have then forced earliest hominins to adapt 499 to more open (grassland savanna) environments by acquiring bipedalism—and the rest is history. 500 Several decades after the proposal of this scenario, where do we stand? 501 502 The landscape of East Africa has dramatically changed during the last 10 million years due to 503 tectonic events leading to specific climatic conditions and associated changes in vegetation 504 structure—from mixed tropical forest to more heterogeneous and arid environments than 505 elsewhere in tropical Africa (144, 145). The trend of progressive aridification did not culminate in 506 the predominance of savanna environments until ~2.0 Ma—roughly coinciding with hominin brain 507 size increase and the appearance of Homo erectus—and was punctuated by alternating episodes of 508 extreme humidity and aridity, resulting in a fluctuating extension of forests through time (144, 509 145). Despite ongoing discussions about early hominin paleoenvironments (woodland with forest 510 patches vs. wooded savanna) (146), evidence from Miocene apes (30, 31) supports that the Pan– 511 Homo LCA inhabited some kind of woodland. Therefore, it has been suggested that the Pan–Homo 512 LCA was probably more omnivorous than chimpanzees (ripe fruit specialists) and likely fed both 513 in trees and on the ground (33)—in agreement with isotopic analyses for Ardipithecus ramidus 514 (41). 515 516 Bipedalism would have emerged due to the selection pressures created by the progressive 517 fragmentation of forested habitats and the need for terrestrial travel from one feeding patch to the 518 next. Data on extant ape positional behaviors (Fig. 4) suggest that hominin terrestrial bipedalism 519 originated as a posture rather than a means of travel on the ground (147) or in trees (140). Rose 520 (39) proposed a long process of increasing commitment to bipedality in the transition to more 521 complex open habitats throughout the Plio-Pleistocene, and Potts (148) argued that key stages in 522 hominin evolution may relate to adaptive responses to cope with highly-variable environments. 523 The fossil and archeological records provide a new twist to the order of evolutionary events in 524 early hominin evolution. The remains of Orrorin and Ardipithecus ramidus indicate that habitual 17 525 terrestrial bipedalism, enhanced precision grasping, and loss of canine honing evolved at the dawn 526 of the human lineage well before brain enlargement (9, 33, 89, 93). It was not until later in time 527 (maybe starting with Australopithecus (149), and continuing with Homo), that some pre-existing 528 hand attributes were co-opted for purposive and systematic stone tool making in more 529 encephalized hominins with more advanced cognitive abilities (38, 150). 530 531 The specialization trap 532 That hominins continuously evolved since the Pan–Homo LCA is universally accepted, but the 533 possibility that all living hominoids (including chimpanzees) experienced their own evolutionary 534 histories is sometimes disregarded. Potts (151) suggested that the greater cognitive abilities of 535 great apes originated to continue exploiting fruit supplies from densely forested environments in 536 front of strong environmental variability. Coupled with locomotor adaptations (e.g., vertical 537 climbing, suspension) enabling an efficient navigation through the canopy, this “cognitive trap” 538 would consist of an adaptive feedback loop between diet, locomotion, cognition and life-history. 539 Although hominids originated approximately during the “Mid-Miocene Climatic Optimum” (~17– 540 15 Ma), their subsequent radiation from ~14 Ma onward paralleled a trend of climatic 541 “deterioration” during the rest of the Miocene (152). Great apes might have initially thrived by 542 evolving particular adaptations to more efficient exploit their habitats, thereby occupying new 543 adaptive peaks without abandoning the same area of the adaptive landscape broadly occupied by 544 earlier stem hominoids. Nevertheless, this evolutionary strategy would become unsustainable once 545 a particular paleoenvironmental threshold was surpassed. This could explain the fate of European 546 dryopiths, which survived for some time under suboptimal conditions (despite the progressive 547 trend of cooling and increased seasonality) until they vanished (94). 548 549 The dietary, locomotor and cognitive specializations of late Miocene great apes would have 550 hindered their shift into new adaptive peaks suitable for the more open environments toward the 551 latest Miocene (153). The Miocene planet of the apes became the time of the more generalist Old 552 World monkeys, enabling their survival in a wider variety of seasonal habitats (30, 92, 154). The 553 same specialization trap can explain the delayed retreat of pongines (and hylobatids) to 554 southeastern Asia throughout the Plio-Pleistocene. The highly specialized orangutans remain 555 extant, but not for long as their habitat continues to shrink. African apes could partially overcome 18 556 the specialization trap by evolving (perhaps in parallel) semiterrestrial adaptations—knuckle 557 walking. Gorillas also expanded their dietary range (more folivorous) and enlarged their body size. 558 Contrary to the view that gorillas are “enlarged” chimpanzees, morphometric analyses indicate 559 that gorillas underwent their own evolutionary history, resulting in different ontogenetic 560 trajectories (155, 156) and postcranial differences that cannot be explained by size-scaling effects 561 (9, 143). Why, when, and how many times knuckle walking evolved is more difficult to explain 562 than the origin of hominin bipedalism. Habitat fragmentation coupled with a higher reliance on 563 arboreal feeding might be invoked (i.e., knuckle walking serves both terrestrial and arboreal 564 locomotion). This idea is difficult to reconcile with the premise that continuous-canopy forests 565 covered the tropical belt of central and western Africa since the Miocene—unless gorillas and 566 chimpanzees evolved in less densely-forested habitats (30, 31, 114), and retreated to tropical 567 forests when outcompeted by hominins and/or cercopithecoids. Ironically, the same specializations 568 that allowed great apes to survive despite major environmental challenges since the late Miocene 569 might ultimately doom them to extinction. 570 571 Hominins might have escaped the great ape specialization trap by evolving novel and more radical 572 adaptations: bipedalism (another specialized orthograde locomotion), concomitant freeing of the 573 hands, and subsequent enhanced manual dexterity, brain configuration, sociosexual behavior, and 574 culturally-mediated technology. Human evolution also reflects the progressive adaptation 575 (biological first, cultural later) to ever-changing environments (39, 148). Some essential changes 576 (upright posture, enhanced cognition) are just the continuation of a trend started in Miocene 577 hominoids (19, 36, 151). While escaping from the great ape specialization trap humans might have 578 fallen in another evolutionary cul-de-sac—with current human activities and overpopulation 579 leading the biosphere to a point beyond return (157). Will humans escape their own specialization 580 trap? 581 582 583 Conclusions and perspectives 584 Fossils uniquely inform deep-time evolutionary studies, which is essential to plan for the future 585 (158). However, we must be aware of the many existing limitations, and the gaps in our knowledge. 586 For example, we need more fossils because we are likely missing vastly more than what we have. 19 587 More fieldwork is necessary to find fossil apes close to the gorilla or chimpanzee lineages, and it 588 is essential to extend such efforts to unexplored or undersampled areas (Fig. 1). It is also essential 589 to continue developing tools of phylogenetic inference. Bayesian approaches are promising, but 590 uncertainty remains about their applicability to morphological data (159). Improvements in the 591 treatment of continuous characters and recent methodological advances for analyzing 3D 592 geometric morphometric data within a cladistic framework (in combination with traditional 593 characters) are promising for reconstructing fossil hominoid phylogeny (160). The oldest (recently 594 retrieved) ancient DNA is ~1 Ma (161). Paleoproteomics could be a complementary solution since 595 it has enabled sampling further back in time up to ~2 Ma, recently confirming the pongine status 596 of Gigantopithecus (162). Future technological advances in paleoproteomics could potentially 597 help to answer key questions by retrieving paleoproteomes from Miocene apes. 598 599 Locomotor reconstructions of the Pan–Homo LCA and other fossil hominoids are seriously 600 hampered by the lack of current analogs. Washburn spotted the fundamental limitation: “it is not 601 possible to bring the past into the laboratory. No one can see a walking Australopithecus” (163, p. 602 67). Such inferences rely on morphofunctional assumptions of bone, joint, or muscle function, but 603 experimentally-derived biomechanical data are required to test these assumptions and provide 604 reliable inferences from fossils. Technological advances now facilitate non-invasive kinematic 605 data collection from animals in their natural environments (164). In turn, experimental and 606 morphological information should be integrated to better predict the locomotion of fossil 607 hominoids. Forward dynamic simulations offer a powerful pathway for predicting de novo 608 movements in fossil species while iterating possible effects of morphology and soft tissue (165). 609 610 Humans are storytellers: Theories of human evolution often resemble “anthropogenic narratives” 611 that borrow the structure of a hero’s journey to explain essential aspects such as the origins of erect 612 posture, the freeing of the hands, or brain enlargement (166). Intriguingly, such narratives have 613 not drastically changed since Darwin (166). We must be aware of confirmation biases and ad hoc 614 interpretations by researchers aiming to confer their new fossil the starring role within a pre- 615 existing narrative. Evolutionary scenarios are appealing because they provide plausible 616 explanations based on current knowledge, but unless grounded in testable hypotheses, they are no 617 more than “just-so stories” (167). 20 618 619 Many uncertainties persist about fossil apes, and the day in which the paleobiology of extinct 620 species can be undisputedly reconstructed is still far. However, current disagreements regarding 621 ape and human evolution would be much more informed if—together with early hominins and 622 living apes—Miocene apes were also included in the equation. This approach will allow us to 623 better discern primitive and derived traits, the common from the specific, or the unique. This is the 624 role of fossil apes in human evolution. 625 626 627 Table 628 Table 1. Simplified taxonomy of extant primates. 629 Order Primates 630 Suborder Strepsirrhini (non-tarsier “prosimians:” lemurs, galagos and lorises) 631 Suborder Haplorrhini (tarsiers and simians) 632 Infraorder Tarsiiformes (tarsiers) 633 Infraorder Simiiformes [=Anthropoidea] (simians or anthropoids: monkeys, apes and 634 humans) 635 Parvorder Platyrrhini (New World monkeys) 636 Parvorder Catarrhini (Old World simians) 637 Superfamily Cercopithecoidea (Old World monkeys) 638 Superfamily Hominoidea (apes and humans) 639 Family Hylobatidae (“lesser apes:” gibbons and siamangs) 640 Family Hominidae (“great apes” and humans) 641 Subfamily Ponginae (the orangutan lineage) 642 Genus Pongo (orangutans) 643 Subfamily Homininae (the African ape and human lineage) 644 Tribe Gorillini (the gorilla lineage) 645 Genus Gorilla (gorillas) 646 Tribe Panini (the chimpanzee lineage) 647 Genus Pan (common chimpanzees and bonobos) 648 Tribe Hominini (the humans linage) 21 649 Genus Homo (humans) 650 651 The adjectives “lesser” and “great” refer to the smaller size of the former relative to great apes and 652 human group, not to old evolutionary notions based on the Scala Naturae. Given that some apes 653 are more closely related to humans than to other apes, the word “ape” is a gradistic term used here 654 informally to refer to all non-hominin hominoids. Finally, the taxonomic convention used (the 655 most common), does not reflect that panins and hominins are monophyletic [although some do; 656 e.g., (168)]. 657 22 658 Figure captions 659 660 661 Fig. 1. Extant and fossil ape distribution. Extant apes live in (or nearby) densely-forested areas 662 around the equator in Africa and Southeast Asia. Except for the recently-recognized tapanuli 663 orangutan (which may represents a subspecies of the Sumatran orangutan), each of the three extant 664 great ape genera presently has two geographically separated species. The Congo River 665 (highlighted) acts as the current barrier between common chimpanzees (Pan troglodytes) and 666 bonobos (P. paniscus). Red stars indicate regions with Miocene sediments (spanning ~23–5.3 667 million years ago) where fossil apes have been uncovered (some regions may contain more than 668 one site; contiguous regions are indicated with different stars if they extend over more than a 669 political region). It is possible that modern great ape habitats do not represent the ancestral 670 environments where the great ape and human clade evolved. Paleontologically, the vast majority 671 of Africa, west of the Rift Valley, remains highly unexplored. Extant ape ranges were taken from 672 the International Union for Conservation of Nature (IUCN Red List). 673 23 674 675 Fig. 2. Pronograde vs. orthograde body plan. (A) Macaque (above) and chimpanzee (below) in 676 typical postures, showing general differences between pronograde and orthograde body plan 677 characteristics. In comparison to a pronograde monkey, the modern hominoid orthograde body 678 plan is characterized by the lack of an external tail (the coccyx being its vestigial remnant), a 679 ribcage that is mediolaterally broad and dorsoventrally shallow, dorsally-placed scapulae that are 680 cranially elevated and oriented, a shorter lower back and long iliac blades. Modern hominoids have 681 higher ranges of joint mobility, such as the full elbow extension shown here, facilitated by a short 682 ulnar olecranon process. The inset further shows differences in lumbar vertebral anatomy, 683 including more dorsally situated and oriented transverse processes in orthograde hominoids. (B) 684 Representatives of each extant hominoid lineage (left column) show different postural variations 685 associated with an orthograde body plan. The orthograde body plan facilitates bipedal walking in 24 686 modern humans, and different combinations of arboreal climbing and below-branch suspension. 687 Knuckle walking in highly terrestrial African apes is seen as a compromise positional behavior 688 superimposed onto an orthograde ape with long forelimbs relative to the hindlimbs. Associated 689 skeletons of fossil hominoids (right column) show that an orthograde body can be disassociated 690 from specific adaptions for suspension (e.g., Pierolapithecus exhibits shorter and less curved digits 691 than Hispanopithecus). Other fossil apes exhibit primitive “monkey-like” pronograde body plans 692 with somewhat more modern ape-like forelimbs (e.g., Nacholapithecus). Approximate age in 693 millions of years ago (Ma) is given to representative fossils of each extinct genus: Ardipithecus 694 (ARA-VP-6/500), Nacholapithecus (KNM-BG35250), Pierolapithecus (IPS21350), 695 Hispanopithecus (IPS18800), and Oreopithecus (IGF 11778). Silhouettes of extant and fossil 696 skeletons are shown at about the same scale. 697 25 698 26 699 Fig. 3. Phylogenetic relationships among living hominoids and chronostratigraphic ranges 700 of fossil hominoids. A time-calibrated phylogenetic tree of living hominoids is depicted next to 701 the spatiotemporal ranges of the fossil hominoids mentioned in the text. Fossil taxa are color-coded 702 based on possible phylogenetic hypotheses. The vertical dotted line indicates that there is a 703 continuity in the African fossil ape record. However, currently, it is sparse between ~14–10 million 704 years ago. Robust and lasting phylogenetic inferences of apes are difficult, in part, due to the 705 fragmentary nature of the fossil record and probable high levels of homoplasy. Many Miocene ape 706 taxa are represented only by fragmentary dentognathic fossils, and the utility of mandibles and 707 molars for inferring phylogeny in apes has been questioned. Another area of uncertainty relates to 708 the position of many early and middle Miocene African apes relative to the crown hominoid node. 709 The discovery or recognition of more complete early Miocene fossil hylobatids would help resolve 710 their position, and thus, what really defines the great ape and human family. Splitting times are 711 based on the molecular clock estimates of Springer et al. (169) (hominoids and hominids) and 712 Moorjani et al. (4), which are more updated for hominines and Pan–Homo. Silhouettes are not to 713 scale. 714 27 715 716 Fig. 4. The positional repertoire preceding human bipedalism. Although one particular 717 behavior can dominate the locomotor repertoire of a given species, the full positional repertoire 718 (postural and locomotor behaviors) of living primates is diverse, complex and not fully understood. 719 For example, some locomotor behaviors are not totally comparable (e.g., monkey quadrupedalism 720 vs. African ape knuckle walking). Furthermore, comprehensive data are not yet available for some 721 extant hominoids (e.g., Gorilla). Bipedalism did not appear de novo in hominins, it existed as a 722 posture or locomotion within a broader Miocene ape positional repertoire. The combined evidence 723 of Miocene apes and early hominins indicate that the locomotor repertoire of the Pan–Homo last 724 common ancestor likely included a combination of positional behaviors not represented among 725 living primates. Over time, bipedal behaviors became the predominant activity within the 726 repertoire of early hominins (and knuckle walking in the chimpanzee lineage). Locomotor 727 behaviors (plus bipedal standing) in each taxon represent percentages of total positional behavior 728 repertoire (full repertoire not shown, hence these do not add to 100%). Data were taken from (Hunt 28 729 2016). Quadrupedalism includes Hunt’s categories “quadrupedal walk” and “quadrupedal run”, 730 suspension includes “suspensory,” “brachiate,” “clamber,” and “transfer.” The locomotor 731 repertoire compositions of the LCA and modern humans (Homo) are conjectural, for illustrative 732 purposes. 733 734 735 References 736 1. C. Darwin, The Descent of Man, and Selection in Relation to Sex (Vol. I). (John Murray, 737 London, 1871). 738 2. T. H. Huxley, Evidence As To Man’s Place in Nature. (Williams and Norgate, London, 739 1863), pp. 159. 740 3. E. Lartet, Note sur un grand singe fossile qui se rattache au groupe des singes superieurs. 741 Comptes Rendus Hebdomadaires des Seances de l'Academie des Sciences 43, 219-223 742 (1856). 743 4. P. Moorjani, C. E. G. Amorim, P. F. Arndt, M. Przeworski, Variation in the molecular 744 clock of primates. P. Natl Acad. Sci. USA 113, 10607-10612 (2016). 745 5. J. Prado-Martinez et al., Great ape genetic diversity and population history. Nature 499, 746 471-475 (2013). 747 6. K. D. Hunt et al., Standardized descriptions of primate locomotor and postural modes. 748 Primates 37, 363-387 (1996). 749 7. J. Stern, Before bipedality. Yearb. Phys. Anthropol. 19, 59-68 (1975). 750 8. S. G. Larson, Parallel evolution in the hominoid trunk and forelimb. Evol. Anthropol. 6, 751 87-99 (1998). 752 9. S. Almécija, J. B. Smaers, W. L. Jungers, The evolution of human and ape hand 753 proportions. Nat. Commun. 6, 7717 (2015). 754 10. M. Nakatsukasa, S. Almécija, D. R. Begun, in The Evolution of the Primate Hand: 755 Anatomical, Developmental, Functional, and Paleontological Evidence, L. T. Kivell, P. 756 Lemelin, G. B. Richmond, D. Schmitt, Eds. (Springer New York, New York, NY, 2016), 757 pp. 485-514. 758 11. A. Keith, Hunterian lectures on man’s posture: Its evolution and disorders. Lecture II. The 759 evolution of the orthograde spine. The British Medical Journal 1, 499-502 (1923). 29 760 12. D. J. Morton, Evolution of man’s erect posture (preliminary report). J. Morphol. 43, 147- 761 179 (1926). 762 13. M. Cartmill, in Functional Vertebrate Morphology, M. Hildebrand, D. Bramble, K. Liem, 763 D. Wake, Eds. (Belknap Press, Cambridge, 1985), pp. 73-88. 764 14. D. R. Pilbeam, D. E. Lieberman, in Chimpanzees and Human Evolution, M. N. Muller, R. 765 W. Wrangham, D. R. Pilbeam, Eds. (The Belknap Press of Harvard University Press, 766 London, England, 2017), pp. 22-141. 767 15. B. G. Richmond, D. R. Begun, D. S. Strait, Origin of human bipedalism: the knuckle- 768 walking hypothesis revisited. Yearb. Phys. Anthropol. 44, 70-105 (2001). 769 16. R. Wrangham, D. Pilbeam, in All apes great and small, B. M. F. Galdikas, N. E. Briggs, 770 L. K. Sheeran, G. L. Shapiro, J. Goodall, Eds. (Kluwer Academic/Plenum Publishers, New 771 York, 2001), pp. 5-17. 772 17. Y. Coppens, East side story: the origin of humankind. Scientific American 270, 62-69 773 (1994). 774 18. D. Pilbeam, N. Young, Hominoid evolution: synthesizing disparate data. C. R. Palevol 3, 775 305-321 (2004). 776 19. S. Moyà-Solà, M. Köhler, D. M. Alba, I. Casanovas-Vilar, J. Galindo, Pierolapithecus 777 catalaunicus, a new Middle Miocene great ape from Spain. Science 306, 1339-1344 778 (2004). 779 20. M. Brunet et al., The first australopithecine 2,500 kilometres west of the Rift Valley 780 (Chad). Nature 378, 273-275 (1995). 781 21. M. Brunet et al., A new hominid from the Upper Miocene of Chad, Central Africa. Nature 782 418, 145-151 (2002). 783 22. G. C. Conroy, M. Pickford, B. Senut, J. Van Couvering, P. Mein, Otavipithecus 784 namibiensis, first Miocene hominoid from southern Africa. Nature 356, 144-148 (1992). 785 23. M. Pickford, Y. Coppens, B. Senut, J. Morales, J. Braga, Late Miocene hominoid from 786 Niger. C. R. Palevol 8, 413-425 (2009). 787 24. B. Senut, M. Pickford, D. Wessels, Panafrican distribution of Lower Miocene Hominoidea. 788 C. R. Acad. Sci. Paris 325, 741-746 (1997). 789 25. P. J. Andrews, L. Martin, The phyletic position of the Ad Dabtiyah hominoid. Bulletin of 790 the British Museum (Natural History). Geology. 41, 383-393 (1987). 30 791 26. G. Suwa, R. T. Kono, S. Katoh, B. Asfaw, Y. Beyene, A new species of great ape from the 792 late Miocene epoch in Ethiopia. Nature 448, 921-924 (2007). 793 27. Y. Kunimatsu et al., A new Late Miocene great ape from Kenya and its implications for 794 the origin of African great apes and humans. P. Natl Acad. Sci. USA 104, 19220-19225 795 (2007). 796 28. D. R. Begun, M. C. Nargolwalla, L. Kordos, European Miocene hominids and the origin 797 of the African ape and human clade. Evol. Anthropol. 21, 10-23 (2012). 798 29. J. Fuss, N. Spassov, D. R. Begun, M. Böhme, Potential hominin affinities of 799 Graecopithecus from the Late Miocene of Europe. PLoS ONE 12, e0177127 (2017). 800 30. P. Andrews, An Ape’s View of Human Evolution. (Cambridge University Press, 801 Cambridge, 2016). 802 31. P. Andrews, Last common ancestor of apes and humans: Morphology and environment. 803 Folia Primatol. 91, 122-148 (2019). 804 32. T. E. Cerling et al., Comment on the paleoenvironment of Ardipithecus ramidus. Science 805 328, 1105-d- (2010). 806 33. T. D. White et al., Ardipithecus ramidus and the paleobiology of early hominids. Science 807 326, 64-86 (2009). 808 34. C. Linnaeus, Systema Naturae per Regna Tria Naturae, Secundum Classes, Ordines, 809 Genera, Species, cum Characteribus, Differentiis. Synonymis, Locis. (Laurentius Salvius, 810 Holmiae [Stockholm], 1758), vol. Tomus I. Editio Decima, reformata 811 35. R. D. Martin, Primate Origins and Evolution: A Phylogenetic Reconstruction. (Chapman 812 and Hall London, 1990). 813 36. D. M. Alba, Cognitive inferences in fossil apes (Primates: Hominoidea): does 814 encephalization reflect intelligence? Journal of Anthropological Sciences 88, 11-48 815 (2010). 816 37. M. Cartmill, Rethinking primate origins. Science 184, 436-443 (1974). 817 38. S. Almécija, C. C. Sherwood, in Evolution of Nervous Systems (2nd Ed), J. Kaas, Ed. 818 (Elsevier, 2017), vol. 3. The Nervous Systems of Non-Human Primates, pp. 299-315. 819 39. M. Rose, The process of bipedalization in hominids. Origine (s) de la bipédie chez les 820 hominidés. CNRS, Paris, 37-48 (1991). 31 821 40. J. M. Plavcan, C. P. van Schaik, P. M. Kappeler, Competition, coalitions and canine size 822 in primates. J. Hum. Evol. 28, 245-276 (1995). 823 41. G. Suwa et al., Paleobiological implications of the Ardipithecus ramidus dentition. Science 824 326, 69-99 (2009). 825 42. M. Pickford, B. Senut, D. Gommery, J. Treil, Bipedalism in Orrorin tugenensis revealed 826 by its femora. C. R. Palevol 1, 191-203 (2002). 827 43. S. L. Washburn, Tools and human evolution. Scientific American 203, 62-75 (1960). 828 44. C. O. Lovejoy, The origin of man. Science 211, 341-350 (1981). 829 45. C. Darwin, On the Origin of Species. Or the Preservation of Favoured Races in the 830 Struggle for Life. (John Murray, London, 1859), pp. 502. 831 46. E. Haeckel, Natürliche Schöpfungsgeschichte. (Georg Reimer, Berlin, 1868). 832 47. E. Dubois, On Pithecanthropus erectus: A transitional form between man and the apes. 833 The Journal of the Anthropological Institute of Great Britain and Ireland 25, 240-255 834 (1896). 835 48. R. A. Dart, Australopithecus africanus: The man-ape of South Africa. Nature 115, 195- 836 199 (1925). 837 49. J. S. Weiner, K. P. Oakley, W. E. Le Gros Clark, The solution of the Piltdown problem. 838 Bulletin of the British Museum (Natural History), Geology 2, 139-146 (1953). 839 50. G. E. Lewis, Preliminary notice of new man-like apes from India. American Journal of 840 Science s5-27, 161-181 (1934). 841 51. H. F. Osborn, The discovery of Tertiary man. Science 71, 1-7 (1930). 842 52. W. L. Straus, Jr., The riddle of man’s ancestry. Q. Rev. Biol. 24, 200-223 (1949). 843 53. W. K. Gregory, How near is the relationship of man to the chimpanzee-gorilla stock? Q. 844 Rev. Biol. 2, 549-560 (1927). 845 54. A. H. Schultz, The skeleton of the trunk and limbs of higher primates. Hum. Biol. 2, 303- 846 438 (1930). 847 55. G. Elliot Smith, The Evolution of Man: Essays. (Oxford University Press, Oxford, 1924). 848 56. H. Weinert, Ursprung der Menschheit. Ueber den engeren Anschluss des 849 Menschengeschlechts an die Menschenaffen. . (1932). 850 57. A. H. Schultz, Characters common to higher primates and characters specific for man. Q. 851 Rev. Biol. 11, 259-283 (1936). 32 852 58. L. S. B. Leakey, A new fossil skull from Olduvai. Nature 184, 491-493 (1959). 853 59. L. S. B. Leakey, P. V. Tobias, J. R. Napier, A new species of the genus Homo from Olduvai 854 Gorge. Nature 202, 7-9 (1964). 855 60. D. C. Johanson, M. Taieb, Plio-Pleistocene hominid discoveries in Hadar, Ethiopia. Nature 856 260, 293-297 (1976). 857 61. M. D. Leakey, R. L. Hay, Pliocene footprints in the Laetoli Beds at Laetoli, northern 858 Tanzania. Nature 278, 317-323 (1979). 859 62. E. L. Simons, D. R. Pilbeam, Preliminary revision of the Dryopithecinae (Pongidae, 860 Anthropoidea). Folia Primatol. 3, 81-152 (1965). 861 63. E. L. Simons, The phyletic position of Ramapithecus. Postilla 57, 1-9 (1961). 862 64. L. S. B. Leakey, An early Miocene member of Hominidae. Nature 213, 155-163 (1967). 863 65. E. Zuckerkandl, R. T. Jones, L. Pauling, A comparison of animal hemoglobins by tryptic 864 peptide pattern analysis. P. Natl Acad. Sci. USA 46, 1349 (1960). 865 66. M. Goodman, Immunochemistry of the primates and primate evolution. Annals of the New 866 York Academy of Sciences 102, 219-234 (1962). 867 67. V. M. Sarich, A. C. Wilson, Immunological time scale for hominid evolution. Science 158, 868 1200-1203 (1967). 869 68. S. L. Washburn, Behaviour and the origin of man. Proceedings of the Royal 870 Anthropological Institute of Great Britain and Ireland, 21-27 (1967). 871 69. L. S. B. Leakey, The relationship of African apes, man, and Old World monkeys. P. Natl 872 Acad. Sci. USA 67, 746-748 (1970). 873 70. W. Hennig, Phylogenetic Systematics. (University of Illinois Press, Urbana, Chicago, 874 London, 1966). 875 71. E. Delson, N. Eldredge, I. Tattersall, Reconstruction of hominid phylogeny: A testable 876 framework based on cladistic analysis. J. Hum. Evol. 6, 263-278 (1977). 877 72. J. Yunis, O. Prakash, The origin of man: a chromosomal pictorial legacy. Science 215, 878 1525-1530 (1982). 879 73. C. G. Sibley, J. E. Ahlquist, The phylogeny of the hominoid primates, as indicated by 880 DNA-DNA hybridization. Journal of Molecular Evolution 20, 2-15 (1984). 881 74. M. Goodman, G. Braunitzer, A. Stangl, B. Schrank, Evidence on human origins from 882 haemoglobins of African apes. Nature 303, 546-548 (1983). 33 883 75. P. Andrews, L. Martin, Cladistic relationships of extant and fossil hominoids. J. Hum. Evol. 884 16, 101-118 (1987). 885 76. M. Miyamoto, J. Slightom, M. Goodman, Phylogenetic relations of humans and African 886 apes from DNA sequences in the psi eta-globin region. Science 238, 369-373 (1987). 887 77. M. Ruvolo, Molecular phylogeny of the hominoids: inferences from multiple independent 888 DNA sequence data sets. Molecular Biology and Evolution 14, 248-265 (1997). 889 78. A. L. Zihlman, J. E. Cronin, D. L. Cramer, V. M. Sarich, Pygmy chimpanzee as a possible 890 prototype for the common ancestor of humans, chimpanzees and gorillas. Nature 275, 744- 891 746 (1978). 892 79. J. Goodall, Tool-using and aimed throwing in a community of free-living chimpanzees. 893 Nature 201, 1264-1266 (1964). 894 80. R. Tuttle, in Phylogeny of the Primates, W. P. Luckett, F. S. Szalay, Eds. (Springer, 1975), 895 pp. 447-480. 896 81. P. Andrews, T. Harrison, in Interpreting the Past: Essays on Human, Primate, and 897 Mammal Evolution. In Honor of David Pilbeam, D. E. Lieberman, R. J. Smith, J. Kelley, 898 Eds. (Brill Academic, Boston & Leiden, 2005). 899 82. K. P. McNulty, Apes and tricksters: The evolution and diversification of humans’ closest 900 relatives. Evolution: Education and Outreach 3, 322-332 (2010). 901 83. D. R. Begun, The Real Planet of the Apes: A New Story of Human Origins. (Princeton 902 University Press, New Jersey, 2016). 903 84. D. Pilbeam, New hominoid skull material from the Miocene of Pakistan. Nature 295, 232- 904 234 (1982). 905 85. P. Andrews, J. E. Cronin, The relationships of Sivapithecus and Ramapithecus and the 906 evolution of the orang-utan. Nature 297, 541-546 (1982). 907 86. D. R. Pilbeam, M. D. Rose, J. C. Barry, S. M. I. Shah, New Sivapithecus humeri from 908 Pakistan and the relationship of Sivapithecus and Pongo. Nature 348, 237-239 (1990). 909 87. B. A. Williams, Comparing levels of homoplasy in the primate skeleton. J. Hum. Evol. 52, 910 480-489 (2007). 911 88. G. E. Erikson, Brachiation in New World monkeys and in anthropoid apes. Symp. Zool. 912 Soc. Lond. 10, 135-163 (1963). 34 913 89. S. Almécija et al., The femur of Orrorin tugenensis exhibits morphometric affinities with 914 both Miocene apes and later hominins. Nat. Commun. 4, 2888 (2013). 915 90. A. S. Hammond, J. M. Plavcan, C. V. Ward, A validated method for modeling anthropoid 916 hip abduction in silico. Am. J. Phys. Anthropol. 160, 529-548 (2016). 917 91. P. Andrews, C. P. Goves, Gibbons and brachiation. Gibbon and Siamang 4, 167-218 918 (1976). 919 92. K. D. Hunt, Why are there apes? Evidence for the co-evolution of ape and monkey 920 ecomorphology. J. Anat. 228, 630-685 (2016). 921 93. C. O. Lovejoy, G. Suwa, S. W. Simpson, J. H. Matternes, T. D. White, The great divides: 922 Ardipithecus ramidus reveals the postcrania of our last common ancestors with African 923 apes. Science 326, 73-106 (2009). 924 94. D. M. Alba, Fossil apes from the Vallès-Penedès Basin. Evol. Anthropol. 21, 254-269 925 (2012). 926 95. M. Dainton, G. A. Macho, Did knuckle walking evolve twice? J. Hum. Evol. 36, 171-194 927 (1999). 928 96. T. Harrison, Apes among the tangled branches of human origins. Science 327, 532-534 929 (2010). 930 97. K. D. Pugh, The Phylogenetic Relationships of Middle-Late Miocene Apes: Implications 931 for Early Human Evolution. The Graduate Center, City University of New York, New York, 932 NY, (2020). 933 98. Y. Chaimanee, V. Lazzari, K. Chaivanich, J.-J. Jaeger, First maxilla of a late Miocene 934 hominid from Thailand and the evolution of pongine derived characters. J. Hum. Evol. 134, 935 102636 (2019). 936 99. J. Kelley, F. Gao, Juvenile hominoid cranium from the late Miocene of southern China and 937 hominoid diversity in Asia. P. Natl Acad. Sci. USA 109, 6882-6885 (2012). 938 100. D. L. Gebo et al., A hominoid genus from the early Miocene of Uganda. Science 276, 401- 939 404 (1997). 940 101. M. Pickford, P. Mein, Early Middle Miocene mammals from Moroto II, Uganda. Beiträge 941 der Paläontologie 30, 361-386 (2006). 942 102. L. MacLatchy, The oldest ape. Evol. Anthropol. 13, 90-103 (2004). 35 943 103. M. Nakatsukasa, Y. Kunimatsu, Nacholapithecus and its importance for understanding 944 hominoid evolution. Evol. Anthropol. 18, 103-119 (2009). 945 104. M. Böhme et al., A new Miocene ape and locomotion in the ancestor of great apes and 946 humans. Nature 575, 489-493 (2019). 947 105. S. Moyà-Solà, M. Köhler, A Dryopithecus skeleton and the origins of great-ape 948 locomotion. Nature 379, 156-159 (1996). 949 106. S. Moyà-Solà et al., First partial face and upper dentition of the Middle Miocene hominoid 950 Dryopithecus fontani from abocador de Can Mata (Vallès-Penedès Basin, Catalonia, NE 951 Spain): taxonomic and phylogenetic implications. Am. J. Phys. Anthropol. 139, 126-145 952 (2009). 953 107. B. R. Benefit, M. L. McCrossin, Miocene hominoids and hominid origins. Annual Review 954 of Anthropology 24, 237-256 (1995). 955 108. D. Pilbeam, Genetic and morphological records of the Hominoidea and hominid origins: a 956 synthesis. Molecular Phylogenetics and Evolution 5, 155-168 (1996). 957 109. D. M. Alba et al., Miocene small-bodied ape from Eurasia sheds light on hominoid 958 evolution. Science 350, aab2625-2621-aab2625-2611 (2015). 959 110. L. de Bonis, G. Bouvrain, D. Geraad, G. Koufos, New hominid skull material from the late 960 Miocene of Macedonia in northern Greece. Nature 345, 712-714 (1990). 961 111. B. Senut et al., First hominid from the Miocene (Lukeino Formation, Kenya). C. R. Acad. 962 Sci. Paris 332, 137-144 (2001). 963 112. B. Wood, T. Harrison, The evolutionary context of the first hominins. Nature 470, 347- 964 352 (2011). 965 113. M. Pickford, B. Senut, Hominoid teeth with chimpanzee-and gorilla-like features from the 966 Miocene of Kenya: implications for the chronology of ape-human divergence and 967 biogeography of Miocene hominoids. Anthropological Science 113, 95-102 (2005). 968 114. S. McBrearty, N. G. Jablonski, First fossil chimpanzee. Nature 437, 105-108 (2005). 969 115. J. DeSilva, E. Shoreman, L. MacLatchy, A fossil hominoid proximal femur from 970 Kikorongo Crater, southwestern Uganda. J. Hum. Evol. 50, 687-695 (2006). 971 116. H. Ishida, M. Pickford, H. Nakaya, Y. Nakano, Fossil anthropoids from Nachola and 972 Samburu hills, Samburu district, Kenya. African study monographs. Supplementary issue. 973 2, 73-85 (1984). 36 974 117. D. R. Begun, in Hominoid Evolution and Climatic Change in Europe, Vol. 2. Phylogeny of 975 the Neogene Hominoid Primates of Eurasia, L. de Bonis, G. D. Koufos, P. Andrews, Eds. 976 (Cambridge University Press, Cambridge, 2001), pp. 231-253. 977 118. P. Andrews, J. Kelley, Middle Miocene dispersals of apes. Folia Primatol. 78, 328-343 978 (2007). 979 119. S. Moyà-Solà, M. Köhler, Recent discoveries of Dryopithecus shed new light on evolution 980 of great apes. Nature 365, 543-545 (1993). 981 120. S. M. Cote, Origins of the African hominoids: an assessment of the palaeobiogeographical 982 evidence. C. R. Palevol 3, 323-340 (2004). 983 121. W. E. Le Gros Clark, The Fossil Evidence for Human Evolution. (University of Chicago 984 Press, Chicago, 1964). 985 122. J. T. J. Stern, Climbing to the top: a personal memoir of Australopithecus afarensis. Evol. 986 Anthropol. 9, 113-133 (2000). 987 123. C. V. Ward, Interpreting the posture and locomotion of Australopithecus afarensis: where 988 do we stand? Yearb. Phys. Anthropol. 45, 185-215 (2002). 989 124. S. A. Williams, T. C. Prang, M. R. Meyer, G. A. Russo, L. J. Shapiro, Reevaluating 990 bipedalism in Danuvius. Nature 586, E1-E3 (2020). 991 125. S. A. Williams, G. A. Russo, Evolution of the hominoid vertebral column: The long and 992 the short of it. Evol. Anthropol. 24, 15-32 (2015). 993 126. C. V. Ward, A. Walker, M. F. Teaford, I. Odhiambo, Partial skeleton of Proconsul nyanzae 994 from Mfangano Island, Kenya. Am. J. Phys. Anthropol. 90, 77-111 (1993). 995 127. E. Delson, An anthropoid enigma: historical introduction to the study of Oreopithecus 996 bambolii. J. Hum. Evol. 15, 523-531 (1986). 997 128. I. Nengo et al., New infant cranium from the African Miocene sheds light on ape evolution. 998 Nature 548, 169-174 (2017). 999 129. A. S. Hammond et al., Insights into the lower torso in late Miocene hominoid Oreopithecus 1000 bambolii. P. Natl Acad. Sci. USA 117, 278-284 (2020). 1001 130. M. Köhler, S. Moyà-Solà, Ape-like or hominid-like? The positional behavior of 1002 Oreopithecus bambolii reconsidered. P. Natl Acad. Sci. USA 94, 11747-11750 (1997). 1003 131. J. D. Currey, Bones: structure and mechanics. (Princeton University Press, 2002). 37 1004 132. C. Ruff, B. Holt, E. Trinkaus, Who’s afraid of the big bad Wolff?: “Wolff’s law” and bone 1005 functional adaptation. Am. J. Phys. Anthropol. 129, 484-498 (2006). 1006 133. T. L. Kivell, A review of trabecular bone functional adaptation: what have we learned from 1007 trabecular analyses in extant hominoids and what can we apply to fossils? J. Anat. 228, 1008 569-594 (2016). 1009 134. B. Demes et al., Patterns of strain in the macaque ulna during functional activity. Am. J. 1010 Phys. Anthropol. 106, 87-100 (1998). 1011 135. L. E. Lanyon, The success and failure of the adaptive response to functional load-bearing 1012 in averting bone fracture. Bone 13, S17-S21 (1992). 1013 136. I. J. Wallace, S. Judex, B. Demes, Effects of load-bearing exercise on skeletal structure and 1014 mechanics differ between outbred populations of mice. Bone 72, 1-8 (2015). 1015 137. J. P. P. Saers, Y. Cazorla-Bak, C. N. Shaw, J. T. Stock, T. M. Ryan, Trabecular bone 1016 structural variation throughout the human lower limb. J. Hum. Evol. 97, 97-108 (2016). 1017 138. Z. Wood et al., Are we crying Wolff? 3D printed replicas of trabecular bone structure 1018 demonstrate higher stiffness and strength during off-axis loading. Bone 127, 635-645 1019 (2019). 1020 139. J. G. Fleagle et al., Climbing: A biomechanical link with brachiation and with bipedalism. 1021 Symposium of the Zoological Society of London 48, 359-375 (1981). 1022 140. S. K. S. Thorpe, R. L. Holder, R. H. Crompton, Origin of human bipedalism as an 1023 adaptation for locomotion on flexible branches. Science 316, 1328-1331 (2007). 1024 141. M. A. Panger, A. S. Brooks, B. G. Richmond, B. Wood, Older than the Oldowan? 1025 Rethinking the emergence of hominin tool use. Evol. Anthropol. 11, 235-245 (2002). 1026 142. E. J. McNutt, B. Zipfel, J. M. DeSilva, The evolution of the human foot. Evol. Anthropol. 1027 27, 197-217 (2018). 1028 143. A. S. Hammond, S. Almécija, Lower ilium evolution in apes and hominins. Anat. Rec. 300, 1029 828-844 (2017). 1030 144. M. A. Maslin et al., East African climate pulses and early human evolution. Quaternary 1031 Science Reviews 101, 1-17 (2014). 1032 145. I. Fer, B. Tietjen, F. Jeltsch, M. H. Trauth, Modelling vegetation change during Late 1033 Cenozoic uplift of the East African plateaus. Palaeogeogr. Palaeoclimatol. Palaeoecol. 1034 467, 120-130 (2017). 38 1035 146. M. Domínguez-Rodrigo, Is the “savanna hypothesis” a dead concept for explaining the 1036 emergence of the earliest hominins? Current Anthropology 55, 59-81 (2014). 1037 147. K. D. Hunt, The evolution of human bipedality: ecology and functional morphology. J. 1038 Hum. Evol. 26, 183-202 (1994). 1039 148. R. Potts, Environmental hypotheses of hominin evolution. Yearb. Phys. Anthropol. 41, 93- 1040 136 (1998). 1041 149. S. Harmand et al., 3.3-million-year-old stone tools from Lomekwi 3, West Turkana, Kenya. 1042 Nature 521, 310-315 (2015). 1043 150. D. M. Alba, S. Moyà-Solà, M. Köhler, Morphological affinities of the Australopithecus 1044 afarensis hand on the basis of manual proportions and relative thumb length. J. Hum. Evol. 1045 44, 225-254 (2003). 1046 151. R. Potts, Paleoenvironmental basis of cognitive evolution in great apes. Am. J. Primatol. 1047 62, 209-228 (2004). 1048 152. J. Zachos, M. Pagani, L. Sloan, E. Thomas, K. Billups, Trends, rhythms, and aberrations 1049 in global climate 65 Ma to present. Science 292, 686-693 (2001). 1050 153. F. Kaya et al., The rise and fall of the Old World savannah fauna and the origins of the 1051 African savannah biome. Nature Ecology & Evolution 2, 241-246 (2018). 1052 154. N. G. Jablonski, M. J. Whitfort, N. Roberts-Smith, X. Qinqi, The influence of life history 1053 and diet on the distribution of catarrhine primates during the Pleistocene in eastern Asia. J. 1054 Hum. Evol. 39, 131-157 (2000). 1055 155. P. Mitteroecker, P. Gunz, M. Bernhard, K. Schaefer, F. L. Bookstein, Comparison of 1056 cranial ontogenetic trajectories among great apes and humans. J. Hum. Evol. 46, 679-698 1057 (2004). 1058 156. S. E. Inouye, Ontogeny and allometry of African ape manual rays. J. Hum. Evol. 23, 107- 1059 138 (1992). 1060 157. A. D. Barnosky et al., Approaching a state shift in Earth’s biosphere. Nature 486, 52-58 1061 (2012). 1062 158. A. D. Barnosky et al., Merging paleobiology with conservation biology to guide the future 1063 of terrestrial ecosystems. Science 355, (2017). 39 1064 159. P. A. Goloboff, A. Torres Galvis, J. S. Arias Becerra, Parsimony and model-based 1065 phylogenetic methods for morphological data: comments on O'Reilly et al. Palaeontology 1066 61, 625-630 (2018). 1067 160. S. A. Catalano, M. D. Ercoli, F. J. Prevosti, The more, the better: The use of multiple 1068 landmark configurations to solve the phylogenetic relationships in musteloids. Syst Biol 1069 64, 294-306 (2015). 1070 161. T. van der Valk et al., Million-year-old DNA sheds light on the genomic history of 1071 mammoths. Nature, (2021). 1072 162. F. Welker et al., Enamel proteome shows that Gigantopithecus was an early diverging 1073 pongine. Nature, (2019). 1074 163. S. Washburn, Human evolution: Science or game. Yearb. Phys. Anthropol. 17, 67-70 1075 (1973). 1076 164. W. I. Sellers, E. Hirasaki, Markerless 3D motion capture for animal locomotion studies. 1077 Biology Open 3, 656-668 (2014). 1078 165. J. A. Nyakatura et al., Reverse-engineering the locomotion of a stem amniote. Nature 565, 1079 351-355 (2019). 1080 166. M. Landau, Narratives of Human Evolution. (Yale University Press, 1993). 1081 167. R. J. Smith, Explanations for adaptations, just-so stories, and limitations on evidence in 1082 evolutionary biology. Evol. Anthropol. 25, 276-287 (2016). 1083 168. M. Goodman et al., Primate evolution at the DNA level and a classification of hominoids. 1084 Journal of Molecular Evolution 30, 260-266 (1990). 1085 169. M. S. Springer et al., Macroevolutionary dynamics and historical biogeography of primate 1086 diversification inferred from a species supermatrix. PLoS ONE 7, e49521 (2012). 1087 1088 1089 Acknowledgments 1090 We are thankful to the many colleagues that motivated and shaped this review through their own 1091 work on the “Miocene ape–hominin transition.” In particular, we would like to highlight the 1092 decades-long work of Peter Andrews, David Begun, Brenda Benefit, Terry Harrison, Bill Jungers, 1093 Jay Kelley, Owen Lovejoy, Laura MacLatchy, Masato Nakatsukasa, Monte McCrossin, Martin 1094 Pickford, David Pilbeam, Brigitte Senut, Jack Stern, Carol Ward, and Tim White. Eric Delson and 40 1095 Santiago Catalano provided constructive criticisms on an earlier version of the manuscript. Kayla 1096 Younkin assisted making Figure 2. Funding: This research has been funded by the Agencia Estatal 1097 de Investigación (CGL2016-76431-P and CGL2017-82654-P, AEI/FEDER EU) and the 1098 Generalitat de Catalunya (CERCA Programme, and consolidated research groups 2017 SGR 86 1099 and 2017 SGR 116 GRC). Author contributions: S.A. designed the review and coordinated the 1100 work; A.S.H. and N.E.T. led the functional morphology discussion; K.D.P., S.M.S and D.M.A. 1101 lead the phylogenetic, evolutionary and paleobiogeographical components. S.A. and N.E.T 1102 designed the figures. All authors discussed and wrote the paper. Competing interests: The authors 1103 declare no competing interests. 41